assignment_policy() updates the value of the 'memory' key of a task to calculate locality scores.
|
resource_requirements['memory'] += nonlocal_data |
However, if a task does not use any PArray, it doubles unnecessary iteration space of check_resources_availability().
In 1000 independent 0.5s tasks, this iteration increases execution time from 43s to 53s.
The resource requirement dictionary update should be avoided if a task does not use any PArray.
assignment_policy() updates the value of the 'memory' key of a task to calculate locality scores.
Parla.py/parla/task_runtime.py
Line 1671 in cbe1469
However, if a task does not use any PArray, it doubles unnecessary iteration space of check_resources_availability().
In 1000 independent 0.5s tasks, this iteration increases execution time from 43s to 53s.
The resource requirement dictionary update should be avoided if a task does not use any PArray.